Correct me if I'm wrong but all he was whiltleblowing is that OpenAI trained on copyrighted content, which is completely normal and expected although its legality is yet to be determined.